Pith. sign in

REVIEW 5 major objections 6 minor 23 references

Bi-directional Curriculum Learning for Graph Anomaly Detection: Dual Focus on Homogeneity and Heterogeneity

T0 review · 5 major / 6 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read The paper claims that a bi-directional curriculum—training one detector copy from the most typical nodes and another from the most atypical—plugged into ten existing graph anomaly detectors improves them on seven datasets, with four…

desk verdict A plausible new curriculum-learning recipe for GAD that deserves a serious referee, but the paper needs a random-ordering control and a defined 'none' baseline before the central mechanism claim holds. read the letter →

arxiv 2501.14197 v1 pith:PLQSWN3W submitted 2025-01-24 cs.LG cs.SIstat.ML

Yitong Hao , Enbo He , Yue Zhang , Guisheng Yin This is my paper · ORCID
classification cs.LGcs.SIstat.ML
keywords graphanomalydetectioncurriculumlearningneuralnetworkshomogeneityheterogeneitytrainingschedulerbi-directionaldifficultyscorenodeclassification
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that graph anomaly detection (GAD) models are currently trained without regard for which nodes are easy or hard, and that existing graph curriculum learning methods only exploit homogeneity, which leaves gains on the table. It introduces Bi-directional Curriculum Learning (BCL), which assigns each node a score equal to the L1 distance between its graph-autoencoder representation and the global mean representation, then trains two copies of any GAD model on opposite orderings of that score—one starting with the most typical nodes and the other with the most atypical—and fuses their outputs. The authors report that this plug-and-play module significantly improves ten existing GAD detectors across seven widely used datasets, with GraphSAGE, BWGNN, AMNET, and GHRN reaching AUC and F1 scores of 1 on the Facebook dataset. A reader should care because the contribution is a training schedule, not a new architecture, so if the claim holds it is a cheap and general upgrade for existing anomaly detectors.

What carries the argument

The central object is the Bi-directional Difficulty Score (BDS), defined in Eq. (3) as $BDS(v_i) = \|h_i - \bar{h}\|_1$, the distance from a node's learned representation to the global mean representation of all nodes. This one scalar produces two curricula: ascending order feeds the homogeneity-focused model, which learns typical patterns first, and descending order feeds the heterogeneity-focused model, which learns atypical patterns first. Around BDS, BCL wraps a continuous training scheduler with three pacing functions—linear, root, and geometric—that control how quickly harder nodes are admitted, and a convex fusion step $\alpha \text{Score}_{\text{homo}} + (1-\alpha)\text{Score}_{\text{hete}}$ that combines the two models' outputs. The machinery's role is to translate the familiar idea of easy-to-hard training into the two complementary signals that a GAD model can exploit.

What would settle it

A concrete test would be to hold everything in BCL fixed—the two-model split, the pacing schedule, and the fusion—but replace the BDS ordering with a random permutation of nodes, and compare against the BDS ordering on the same seven datasets. If the random-order control matches BCL's AUC improvements under multiple seeds, the claim that the score-based difficulty measure drives the gain is falsified. A second check would rerun the reported perfect AUC/F1 of 1 on Facebook across many random seeds and fixed hyperparameter grids, since a perfect score on a 1081-node benchmark with only 2.3% anomalies can be brittle.

Watch

Extended reading notes

Core claim

The central claim is that a graph anomaly detector can learn better when its training set is ordered along both directions of a single difficulty spectrum: nodes closest to the global mean representation are easiest for a homogeneity-focused model, and nodes farthest from the mean are easiest for a heterogeneity-focused model. The paper constructs this ordering with the Bi-directional Difficulty Score $BDS(v_i) = \|h_i - \bar{h}\|_1$, where $h_i$ comes from a two-layer GCN graph autoencoder and $\bar{h}$ is the average of all node representations. BCL sorts nodes by BDS ascending and descending, feeds each sorted sequence to an identical copy of the detector through a continuous pacing function that gradually admits harder nodes, and computes the final anomaly score as $\alpha \cdot \text{Score}_{\text{homo}} + (1-\alpha) \cdot \text{Score}_{\text{hete}}$. The paper's experiments, run on seven datasets and ten detectors, are offered as evidence that this dual-direction curriculum consistently outperforms both the unmodified detectors and prior homogeneity-only curriculum methods.

Load-bearing premise

The load-bearing premise is that the distance between a node's learned representation and the global mean representation genuinely orders nodes by learning difficulty for every GAD method—that low-BDS nodes really are the easy ones for the homogeneity direction and high-BDS nodes really are the easy ones for the heterogeneity direction—so the curriculum is not just a fixed arbitrary ordering.

Editorial extensions

If this is right

  • Any existing graph anomaly detector can be upgraded without changing its architecture; the paper demonstrates this for MLP, GCN, GAT, GraphSAGE, and six state-of-the-art detectors.
  • Detectors built to handle heterophily, such as CARE-GNN and Split-GNN, still gain from the heterogeneity-first curriculum, suggesting the two directions capture complementary information rather than duplicating model strengths.
  • The gain is sensitive to the fusion weight $\alpha$; the paper shows different detectors prefer different balances between homogeneity and heterogeneity, so tuning this single parameter is part of the reported improvement.
  • Among the three pacing functions, the linear scheduler gives the most consistent improvements on the Amazon dataset, which the paper recommends as the default choice.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because BDS is computed with an unsupervised graph autoencoder before any label is used, the same node ordering could in principle be applied to unlabeled or semi-supervised anomaly detection tasks, although the paper only evaluates supervised detectors.
  • The Facebook dataset's small size (1081 nodes, 2.3% anomalies) makes perfect scores plausible even without a curriculum, so the strongest test of the claim would be a stress-run on a large, low-anomaly-rate benchmark where small differences in ranking matter.
  • The two-model fusion doubles the detector's forward pass cost; an alternative single-model design with two output heads or epoch-switched ordering might recover most of the gain at half the inference cost, but that is an extension, not a claim in the paper.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 6 minor

Summary. The paper proposes Bi-directional Curriculum Learning (BCL), a plug-and-play training module for graph anomaly detection (GAD). BCL first trains a graph autoencoder and computes a Bi-directional Difficulty Score (BDS) for each node as the L1 distance between the node's embedding and the global mean embedding. Nodes are then sorted in ascending BDS order to form a homogeneity-focused curriculum and in descending order to form a heterogeneity-focused curriculum. Two copies of the base GAD detector are trained with these curricula using a paced subset selection, and their anomaly scores are combined via a weighted fusion. The authors report that BCL improves ten GAD models across seven datasets, including perfect AUC and F1 scores on the Facebook dataset for several detectors, and they perform ablations over pacing functions, the fusion weight, and the curriculum hyperparameters.

Significance. If the empirical claims are correct, BCL would be a simple, model-agnostic way to improve many graph anomaly detectors, which is potentially impactful given the plug-and-play framing. The breadth of the evaluation (ten detectors and seven datasets) is a strength, and the paper addresses an interesting and under-explored question: how to design curricula for GAD when both homogeneous and heterogeneous nodes can be informative. The complexity analysis is reasonable. However, the causal role of the BDS ordering is not established: the paper lacks random-ordering controls, the 'none' column in Table 3 is undefined, and no variance or significance measures are reported. Because the central claim is that the curriculum ordering itself drives the gains, these omissions are load-bearing. The paper currently reads as a promising but unverified empirical proposal rather than a fully supported contribution.

major comments (5)
  1. [Section 4.2, Eq. (7)] The linear pacing function is written as g(t) = min(1, λ0 + (1 - λ0 * t / T)). As written, at t = 0 this gives 1 + λ0, so the function does not start at λ0; the intended formula is presumably λ0 + (1 - λ0) * t / T. Since the paper recommends the linear pacing function in RQ4, this formula error affects the reproducibility of the main recommended configuration.
  2. [Table 3] The 'none' column is never defined. If 'none' denotes the original single-model detector without BCL, then the comparison conflates the curriculum effect with the effect of training two models and fusing their scores; if 'none' denotes a two-model version without pacing, that must be stated explicitly. The gap between the 'none' values and the base detectors in Table 2 (e.g., MLP AUC 0.6705 vs. MLP+BCL 'none' AUC 0.8512 on Amazon) suggests that much of the gain may come from the two-model fusion rather than from the curriculum ordering. Please define this condition and add the appropriate controls.
  3. [Section 4.1 and Algorithm 1] The central claim that BDS-based ordering drives the improvements is not supported by any random-ordering or reversed-ordering ablation. BDS is, by construction, an unsupervised deviation score: Figure 3 itself shows that anomalous nodes cluster at high BDS, so the heterogeneity curriculum is essentially training on likely anomalies first. The observed gains could therefore come from (a) the two-model fusion in Eq. (10), (b) BDS acting as an unsupervised anomaly prior, or (c) the specific ascending/descending order. Please report random-order and reversed-order baselines with the same two-model fusion to isolate the effect of the curriculum ordering.
  4. [Section 5, Tables 2 and 3] No variance, error bars, or significance tests are reported. Many of the claimed improvements are very small (for example, several F1 changes are below 0.001), and some BCL results are worse than the corresponding single-direction baselines. Without repeated runs or paired tests, the statement that BCL 'significantly improves' ten detectors is not statistically supported. Please report standard deviations over at least five seeds or provide permutation/paired tests.
  5. [Section 5.2, Facebook results] The claim that 'GraphSAGE, BWGNN, AMNET, and GHRN combined with BCL achieve AUC and F1 scores of 1 on the Facebook dataset' and that 'no previous study has been able to achieve' this is overstated. The Facebook dataset has only 1,081 nodes, and Table 2 shows that the HomoCL baseline alone already achieves 1.0000 AUC and 1.0000 F1 for GraphSAGE, so BCL is not necessary for this perfect result. The statement should be tempered and supported by comparison with previously reported results under identical protocols.
minor comments (6)
  1. [Eq. (3)] The text says '||·||1 denotes the Euclidean norm,' but the subscript 1 denotes the L1 (Manhattan) norm. Please correct the wording.
  2. [Algorithm 1, line 13] The expression Vhomo ← {Vtrain[i] | i ∈ Qhomo[1 : λt × |V|]} is ambiguous because Qhomo contains node identifiers, not positions in Vtrain. Please clarify the indexing.
  3. [Sections 5.5 and 5.6] The figure references are inconsistent: Section 5.5 refers to 'Figure 3' when discussing α, and Section 5.6 refers to 'Figure 4' when discussing λ0 and T. These should be Figure 4 and Figure 5, respectively.
  4. [Section 5.1] There is a stray word 'Avatar' in the Experimental Details paragraph; it appears to be an editing artifact and should be removed.
  5. [Section 5.4 and Table 3] The text uses 'rootedness pacing function' and the table header uses 'geomo'; these should be 'root' and 'geometric' for consistency.
  6. [Overall] No code or detailed reproducibility instructions are provided. Given that the contribution is a plug-and-play training module, releasing code would substantially strengthen the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: BDS is an unsupervised GAE embedding distance, and the reported gains are empirical comparisons against existing detectors.

full rationale

The core difficulty score BDS in Eq. (3) is an unsupervised distance-to-centroid computed from a graph-autoencoder embedding; it is not fit to the GAD labels or to the final anomaly scores. The curriculum ordering is a training schedule, and the final score is a fusion of two supervised detectors via Eq. (10); neither the ordering nor the fusion reduces to BDS by construction. Hyperparameters α, λ0, and T are tuned on a validation grid, which is ordinary hyperparameter selection rather than a fitted parameter renamed as a prediction. No load-bearing self-citation or imported uniqueness theorem appears: prior work such as CLNode and RCL is used only as baselines and motivation. Figure 3 does show that BDS separates anomalous nodes, but that is evidence for the difficulty premise rather than a circular reduction. Potential concerns about causal attribution—such as the undefined 'none' column in Table 3 and the absence of a random-ordering control—bear on whether the curriculum order is the mechanism behind the improvements, not on definitional circularity, so they do not raise the circularity score under the stated rules.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claims rely on three unproven premises: BDS is a meaningful difficulty measure, both homogeneity and heterogeneity directions are useful for GAD, and easy-to-hard scheduling transfers. The numerical results additionally depend on tuned hyperparameters alpha, lambda_0, and T, and on the choice of pacing function. No invented entities are introduced.

free parameters (4)
  • alpha = tuned in {0.1, 0.2, ..., 0.9}
    Balances homogeneous and heterogeneous score contributions in Eq. (10); set per dataset and detector by validation.
  • lambda_0 = tuned in {0.1, 0.2, ..., 0.9}
    Initial fraction of easiest nodes used by the pacing functions; controls curriculum speed.
  • T = 10% to 90% of the original method's training rounds
    Epoch when the pacing function reaches 1; controls how quickly hard nodes are introduced.
  • pacing function = linear selected for most methods
    Choice among linear, root, and geometric affects training order and performance (Table 3).
assumptions (3)
  • ad hoc to paper The distance-to-centroid in a graph autoencoder embedding is a valid difficulty measure for GAD models.
    Section 4.1, Eq. (3), introduces BDS without derivation or theoretical justification.
  • domain assumption GAD models can make use of both homogeneous and heterogeneous easy nodes.
    Section 1 and Section 4.2 use this to justify two curricula; no formal definition is given.
  • domain assumption Training with an easy-to-hard schedule transfers to better anomaly detection.
    This is the standard curriculum learning premise, taken from cited CL work; it is not proven for GAD.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bi-directional Curriculum Learning for Graph Anomaly Detection: Dual Focus on Homogeneity and Heterogeneity." pith.science (2026). https://pith.science/paper/PLQSWN3W

@misc{pith2026250114197,
  author       = {Pith},
  title        = {Pith review of: Bi-directional Curriculum Learning for Graph Anomaly Detection: Dual Focus on Homogeneity and Heterogeneity},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PLQSWN3W}},
  note         = {Machine review of arXiv:2501.14197}
}
read the original abstract

Graph anomaly detection (GAD) aims to identify nodes from a graph that are significantly different from normal patterns. Most previous studies are model-driven, focusing on enhancing the detection effect by improving the model structure. However, these approaches often treat all nodes equally, neglecting the different contributions of various nodes to the training. Therefore, we introduce graph curriculum learning as a simple and effective plug-and-play module to optimize GAD methods. The existing graph curriculum learning mainly focuses on the homogeneity of graphs and treats nodes with high homogeneity as easy nodes. In fact, GAD models can handle not only graph homogeneity but also heterogeneity, which leads to the unsuitability of these existing methods. To address this problem, we propose an innovative Bi-directional Curriculum Learning strategy (BCL), which considers nodes with higher and lower similarity to neighbor nodes as simple nodes in the direction of focusing on homogeneity and focusing on heterogeneity, respectively, and prioritizes their training. Extensive experiments show that BCL can be quickly integrated into existing detection processes and significantly improves the performance of ten GAD anomaly detection models on seven commonly used datasets.

Figures

Figures reproduced from arXiv: 2501.14197 by the authors.

Figure 1
Figure 1. , GAD models not only handle homogeneity but also heterogeneity. Therefore, for GAD models, nodes with both high homogeneity and high heterogeneity should be consid￾ered easy samples. Existing GCL methods overlook this as￾pect, resulting in suboptimal performance when applied to GAD models [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overall framework of BCL The training graph fed into the Graph Autoencoder, and the learning objective can be presented as follows: w(t) pre = arg min W L(f(X;W), X) (1) A two-layer GCN is used as an encoder to aggregate neigh￾borhood information for each node and obtain the i-th node representations hi : hi = GCN(Xi , A;W) (2) Next, the global mean of all node representations h is com￾puted as a baseline, which… view at source ↗
Figure 3
Figure 3. Comparison of Difficult Measure for CLNode and BCL [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: The Impact of Training Scheduler In this section, we investigate how the hyperparameters λ0 and T affect the performance of BCL. Where λ0 controls the initial number of training nodes, while T controls the speed of introducing difficult nodes into the training process.…
Figure 4
Figure 4. Figure 4: The Impact of different α on the performance The parameter α serves to balance the BCL’s focus be￾tween homogeneity and heterogeneity. Specifically, a larger α directs the framework to emphasize homogeneous fea￾tures, whereas a smaller α shifts the focus towards hetero…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 20 canonical work pages

  1. [1]

    Can abnormality be detected by graph neural networks? In Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI),

    [Chai et al., 2022] Ziwei Chai, Siqi You, Yang Yang, Shil- iang Pu, Jiarong Xu, Haoyang Cai, and Weihao Jiang. Can abnormality be detected by graph neural networks? In Proceedings of the 31st International Joint Conference on Artificial Intelligence (IJCAI),

    work page 2022

  2. [8]

    Semi-supervised classification with graph convolutional networks

    [Kipf and Welling, 2016] Thomas Kipf and Max Welling. Semi-supervised classification with graph convolutional networks. ArXiv, abs/1609.02907,

    arXiv 2016

  3. [11]

    Sheng, Hui Xiong, and Le- man Akoglu

    [Ma et al., 2023] Xiaoxiao Ma, Jia Wu, Shan Xue, Jian Yang, Chuan Zhou, Quan Z. Sheng, Hui Xiong, and Le- man Akoglu. A comprehensive survey on graph anomaly detection with deep learning. IEEE Transactions on Knowledge and Data Engineering , 35(12):12012–12038,

    work page 2023

  4. [12]

    Weidele Daniel, Bellei Claudio, Robinson Tom, and E

    [Mark et al., 2019] Weber Mark, Domeniconi Giacomo, Chen Jie, Karl I. Weidele Daniel, Bellei Claudio, Robinson Tom, and E. Leiserson Charles. Anti-money laundering in bitcoin: Experimenting with graph convolutional networks for financial forensics. ArXiv,

    work page 2019

  5. [13]

    Deep graph anomaly detection: A survey and new perspectives

    [Qiao et al., 2024] Hezhe Qiao, Hanghang Tong, Bo An, Ir- win King, Charu Aggarwal, and Guansong Pang. Deep graph anomaly detection: A survey and new perspectives. ArXiv, abs/2409.09957,

    arXiv 2024

  6. [17]

    Rethinking graph neural networks for anomaly de- tection

    [Tang et al., 2022] Jianheng Tang, Jiajin Li, Ziqi Gao, and Jia Li. Rethinking graph neural networks for anomaly de- tection. In International Conference on Machine Learn- ing,

    work page 2022

  7. [18]

    Graph attention networks

    [Veliˇckovi´c et al., 2018] Petar Veliˇckovi´c, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Li `o, and Yoshua Bengio. Graph attention networks. In Interna- tional Conference on Learning Representations,

    2018

  8. [20]

    [Wang et al., 2025] Conghao Wang, Gaurav Asok Kumar, and Jagath C

    Association for Computing Ma- chinery. [Wang et al., 2025] Conghao Wang, Gaurav Asok Kumar, and Jagath C. Rajapakse. Drug discovery and mechanism prediction with explainable graph neural networks. Scien- tific Reports, 15,

    work page 2025

Show all 23 references
  1. [21]

    Clnode: Cur- riculum learning for node classification

    [Wei et al., 2022] Xiaowen Wei, Xiuwen Gong, Yibing Zhan, Bo Du, Yong Luo, and Wenbin Hu. Clnode: Cur- riculum learning for node classification. Proceedings of the Sixteenth ACM International Conference on Web Search and Data Mining,

    2022

  2. [22]

    Splitgnn: Spectral graph neural network for fraud detection against heterophily

    [Wu et al., 2023] Bin Wu, Xinyu Yao, Boyan Zhang, Kuo- Ming Chao, and Yinsheng Li. Splitgnn: Spectral graph neural network for fraud detection against heterophily. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management , CIKM ’23,

    2023

  3. [23]

    Curriculum learning for graph neural net- works: Which edges should we learn first

    [Zhang et al., 2023] Zhengwu Zhang, Junxiang Wang, and Liang Zhao. Curriculum learning for graph neural net- works: Which edges should we learn first. ArXiv, abs/2310.18735, 2023

    2023 arXiv

  4. [1958]

    Predicting dynamic embedding trajectory in temporal interaction networks

    [Srijan et al., 2019] Kumar Srijan, Zhang Xikun, and Leskovec Jure. Predicting dynamic embedding trajectory in temporal interaction networks. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining,

    2019

  5. [2015]

    The perceptron: a probabilistic model for information storage and organiza- tion in the brain

    [Rosenblatt, 1958] Frank Rosenblatt. The perceptron: a probabilistic model for information storage and organiza- tion in the brain. Psychological review, 65 6:386–408,

    1958

  6. [2016]

    A universal adaptive algorithm for graph anomaly detection

    [Li et al., 2025] Yuqi Li, Guosheng Zang, Chunyao Song, and Xiaojie Yuan. A universal adaptive algorithm for graph anomaly detection. Information Processing and Management, 62(1):103905,

    2025

  7. [2017]

    F, and Singaraju Srinivasulu

    [Kavitha et al., 2021] M Kavitha, P V V S Srinivas, P.S.Latha Kalyampudi, Choragudi S. F, and Singaraju Srinivasulu. Machine learning techniques for anomaly de- tection in smart healthcare. In 2021 Third International Conference on Inventive Research in Computing Applica- tion...

    2021

  8. [2018]

    Curgraph: Curriculum learning for graph classification

    [Wang et al., 2021] Yiwei Wang, Wei Wang, Yuxuan Liang, Yujun Cai, and Bryan Hooi. Curgraph: Curriculum learning for graph classification. In Proceedings of the Web Conference 2021, WWW ’21, page 1238–1248, New York, NY , USA,

    2021

  9. [2019]

    Ad- dressing heterophily in graph anomaly detection: A per- spective of graph spectrum

    [Gao et al., 2023] Yuan Gao, Xiang Wang, Xiangnan He, Zhenguang Liu, Huamin Feng, and Yongdong Zhang. Ad- dressing heterophily in graph anomaly detection: A per- spective of graph spectrum. In WWW,

    2023

  10. [2020]

    Elhadad, Kin Fun Li, and Fayez Gebali

    [Elhadad et al., 2019] Mohamed K. Elhadad, Kin Fun Li, and Fayez Gebali. Fake news detection on social media: A systematic survey. In 2019 IEEE Pacific Rim Confer- ence on Communications, Computers and Signal Process- ing (PACRIM), pages 1–8,

    2019

  11. [2021]

    Enhancing graph neural network-based fraud detectors against camouflaged fraudsters

    [Dou et al., 2020] Yingtong Dou, Zhiwei Liu, Li Sun, Yu- tong Deng, Hao Peng, and Philip S Yu. Enhancing graph neural network-based fraud detectors against camouflaged fraudsters. In Proceedings of the 29th ACM International Conference on Information and Knowledge Management (...

    2020

  12. [2022]

    Cuco: Graph representation with cur- riculum contrastive learning

    [Chu et al., 2021] Guanyi Chu, Xiao Wang, Chuan Shi, and Xunqiang Jiang. Cuco: Graph representation with cur- riculum contrastive learning. In International Joint Con- ference on Artificial Intelligence,

    2021

  13. [2023]

    Hamilton, Rex Ying, and Jure Leskovec

    [Hamilton et al., 2017] William L. Hamilton, Rex Ying, and Jure Leskovec. Inductive representation learning on large graphs. In Proceedings of the 31st International Con- ference on Neural Information Processing Systems , page 1025–1035,

    2017

  14. [2024]

    Collective opinion spam detection: Bridging re- view networks and metadata

    [Rayana and Akoglu, 2015] Shebuti Rayana and Leman Akoglu. Collective opinion spam detection: Bridging re- view networks and metadata. In Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , KDD ’15, page 985–994,

    2015

  15. [2025]

    Pick and choose: A gnn-based imbalanced learning approach for fraud detection

    [Liu et al., 2021] Yang Liu, Xiang Ao, Zidi Qin, Jianfeng Chi, Jinghua Feng, Hao Yang, and Qing He. Pick and choose: A gnn-based imbalanced learning approach for fraud detection. In Proceedings of the Web Conference 2021, pages 3168–3177,

    2021

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.